Question Answering on Truthful-QA (Info/Truth Metrics)
99.2Info AccuracyUSC
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| USCModel=Llama 22025.03 | 99.2 | 39.4 | 39.4 | |
| BoNModel=Llama 22025.03 | 99.2 | 41.1 | 40.4 | |
| BoNModel=Llama 32025.03 | 99 | 40.4 | 40.4 | |
| GreedyModel=Llama 22025.03 | 98.2 | 38.2 | 37.9 | |
| MDModel=Llama 22025.03 | 98 | 44.1 | 44.1 | |
| USCModel=Llama 32025.03 | 97.7 | 42.9 | 42.1 | |
| SRModel=Llama 32025.03 | 97 | 44.8 | 43.8 | |
| MDModel=Gemma 22025.03 | 96.8 | 54.7 | 50.2 | |
| GreedyModel=Llama 32025.03 | 96.5 | 43.6 | 42.4 | |
| IDModel=Llama 32025.03 | 96.5 | 41.9 | 40.9 | |
| FSCModel=Gemma 22025.03 | 96.5 | 49 | 48 | |
| DoLaModel=Llama 32025.03 | 95.5 | 40.6 | 39.2 | |
| MDModel=Llama 32025.03 | 95.3 | 47.1 | 46.1 | |
| USCModel=Gemma 22025.03 | 95 | 40.6 | 39.4 | |
| DoLaModel=Gemma 22025.03 | 94.8 | 40.1 | 38.9 | |
| IDModel=Llama 22025.03 | 94.6 | 40.9 | 38 | |
| BoNModel=Gemma 22025.03 | 94.1 | 45.1 | 45.1 | |
| FSCModel=Llama 32025.03 | 93.8 | 46.5 | 44.3 | |
| GreedyModel=Gemma 22025.03 | 93.3 | 45.5 | 43.6 | |
| SRModel=Gemma 22025.03 | 91.7 | 42.1 | 40.2 | |
| FSCModel=Llama 22025.03 | 91.6 | 39.5 | 38.7 | |
| DoLaModel=Llama 22025.03 | 90.6 | 38.4 | 37.2 | |
| SRModel=Llama 22025.03 | 90.2 | 38.9 | 37.5 | |
| CNTPBackbone=Llama-2-7B-Chat2025.07 | 89.2 | 84.8 | 74 | |
| Stochastic DecodingBackbone=Llama-2-7B-Chat2025.07 | 88 | 78 | 66 | |
| IDModel=Gemma 22025.03 | 81.1 | 46.5 | 36.5 | |
| Greedy DecodingBackbone=Llama-2-7B-Chat2025.07 | 78.5 | 79.1 | 57.6 |